Wiki Statistics
« Hier, j'ai subi, par exemple, l'influence rafraîchissante de quelques pages de statistiques. Si on y réfléchit bien, le mystère de l'Univers se trouve aussi là. Même s'il n'y paraît pas. » — Fernando Pessoa
What are the measurements that we can use to determine the "health" of a wiki community ?
We need a common terminology to be able to measure progress & health of a wiki.
What number of users do you expect on a new wiki and how does that change when a wiki grows ?
Be careful that what you're measuring (and reporting on) doesn't negatively affect the behaviour of the users. Meaurement can kill creative activity.
External links Erik Zachte's WikiStats program, compete.com statsb
Participants
* PatrickMichaud * BrandonCsSanders * AngelaBeesley * DaveCrosby * WindySawczyn * RaymondKing
Measurements
* Number of users making at least 5 edits/month as it evolves through time * Number of characters per edit * Minimum Description Length ... MDL of what the user has done for the month * Word vomitters * How many times was it edited afterward * Measuring every action * How many quality pages do we have
Measuring Every Action
* If we measure every action (Quality and Quantity)
What we measure
Quantity
- MDL to discover a parsimonious (mmh, is this a pun ?) description of activity
- Who ... demographics
- Number of visits
- Number of edits
- Measure of "how much change" MDL (number of characters)
- Number of editors
- Number of registered users (logged in)
- Number of pages in the wiki
- Number of links
- Number of wanted pages (suggested articles)
- Amount of activity on a page
- Amount of activity for a user
- Number of subscribers
- Quantity of Vandalism
- Amount of time to catch an episode of vandalism
- Negative energy (how many reverts)
- Number of "common misspellings"
- Amount of discussion
- Ratio of chitchat to article crunching
- Numbers of various categories (admins, etc)
- Retension rate ... how many people are leaving
- Percentage of 5+ edits folks who haven't logged in for over a month
- What is the absolute value ... what is the turnover
- How many registered users login but don't edit
- When you get the most people coming through
- Anonymous versus logged in users
- Anonymous edits vs logged in edits
- How many are actually reading .. how long do they stay on a page
Quality
- What is quality ?
Targets for the Measures
Cautionary Tales
- Putting edit count on the main page changes behaviour
- People editing for the wrong reason
- Be careful what you advertise
- What you measure and report changes the dynamics of the community
